Choropleth Maps

Geographic map (U.S., world, continent) that shows how a variable is geographically distributed.
R has great packages like tigris (for U.S.) and rnaturalearth (for world maps).
Install and load the packages at the top of the R script.

U.S. Maps Using Tigris

Start simple: Use tigris to get a national map with state boundaries (take out PR)
This code will use tigris to create a data object stored in your environment. Note that adding “shift_geometry” is a key command for putting Alaska and Hawaii on the bottom left.

st <- states(cb = TRUE, resolution = "20m", progress_bar=FALSE) |>
  filter(NAME != "Puerto Rico") |>
  shift_geometry()

U.S. Maps Using Tigris

Check the “class” of this data/maps object. You’ll see it’s a “special features” (sf) and a data frame.
- special features (sf) = contains geographical information for mapping. Check data.

class(st)

[1] "sf"         "data.frame"

Quick Map

ggplot(st) + 
  geom_sf()

Quick Map

Remove gray background with “theme_void”:

ggplot(st) + 
  geom_sf() + 
  theme_void()

Quick Map

Change fill and border colors and borderline thickness:

ggplot(st) + 
  geom_sf(fill="dodgerblue", color="white", linewidth=.1) + 
  theme_void()

Choropleth Map: Biden v. Trump 2020

Import 2020 state election data

v2020 <- read_csv("us_vote_2020.csv")

Data Prep for Choropleth Map

We need to merge our election data (v2020) with our “sf” maps object, “st”
See Ch. 3 in ModernDive, section 3.7 (using the “join” function)

Data Prep for Choropleth Map

First, match state names across datasets; use the “mutate” function, which generates a new variable in the data object.

v2020 <- v2020 |> mutate(NAME=state)

This sets up for the merge, which requires you to have one identification variable in common between the two datasets.
In this case, that’s going to be “NAME.”

Merge Data Into Maps Object

Merge our v2020 data object into our st object containing geographic information using “left_join.”
Put the “master” data object first (i.e., st) and the data object that you’re merging into the master object second.
- Always merge the data object into the simple features (maps) object
“left_join” automatically merges by the variable that is common between the two datasets. Again, in this example, that’s “NAME.”

stv20 <- left_join(st, v2020)

Map It!

Using “theme_map” (from ggthemes), center title, make a little bigger (expand=FALSE).
- Check out some of my other specifications that precede this one.

ggplot(stv20, aes(fill = called)) + 
  geom_sf(color = "gray90", linewidth=.2) + 
  scale_fill_manual(values = c("blue", "red"), labels=c("Biden", "Trump")) + 
  theme_map() + coord_sf(expand=FALSE) +
  labs(fill = "",
       title = " 2020 US presidential election results by state",
       caption = "Note: Nebraska and Maine split electoral college votes by congressional district") + 
  theme(plot.title=element_text(face="bold", size=12, hjust=.5), 
        plot.caption = element_text(size=10), 
        legend.text = element_text(size = 8))

Map It!

Counties, Palettes

From line 232 in R script:

Counties, Palettes

“scale_gradient2” to define midpoint (line 265)

World Maps Using rnaturalearth

Bring in world map from rnaturalearth

w <- ne_countries(scale = "medium", returnclass = "sf")

World Maps Using rnaturalearth

w |> 
  ggplot() + 
  geom_sf(aes(fill=income_grp)) + 
  scale_fill_brewer(palette = "YlGnBu")

World Maps Using rnaturalearth

Change direction of scale - higher values of income darker

w |> 
  ggplot() + 
  geom_sf(aes(fill=income_grp)) + 
  scale_fill_brewer(palette = "YlGnBu", direction=-1)

World Maps Using rnaturalearth

Remove Antarctica and others; remove gray background, relabel legend, theme map
Use “Robinson” projection (NYT standard), crs=‘ESRI:54030’

w |> 
  filter(type != "Dependency", type !="Disputed", type != "Indeterminate") |>
  ggplot() + geom_sf(aes(fill=income_grp), color="black", linewidth=.1) + 
  scale_fill_brewer(palette = "YlGnBu", direction=-1, 
                    labels=c("High OECD", "High non-OECD", "Upper middle", "Lower middle", "Low")) +
  coord_sf(crs='ESRI:54030', expand=FALSE) +
  labs(fill = "Income Level", title = "World Map, Country Income Levels",
       caption="Source: R Natural Earth Data") +
  theme_map() +
  theme(plot.title=element_text(face="bold", size=12, hjust=.5), 
        plot.caption = element_text(size=10), 
        legend.title = element_text(size=10),
        legend.text = element_text(size = 8))

World Maps Using rnaturalearth

Merge in Variable for World Choropleth Map

Let’s look at the degree of political, rhetorical attacks (by politicians and elites) on the judiciary.
Merge in VDem data; I have already created a country id variale (“geounit”) that is common between data objects.

v <- read_dta("vdem.dta")

# Merge vdem with world data
vw <- left_join(w, v)

Generate the Map

Let’s look at the degree of political, rhetorical attacks (by politicians and elites) on the judiciary.
Merge in VDem data; I have already created a country id variale (“geounit”) that is common between data objects.

Generate the Map

vw |>
  filter(type != "Dependency", type !="Disputed", type != "Indeterminate") |>
  ggplot() + 
  geom_sf(color="black", linewidth=0.1, aes(fill = as.factor(v2jupoatck_ord))) + 
  coord_sf(crs='ESRI:54030', expand=FALSE) +
  scale_fill_brewer(type = "seq", palette = "YlGnBu", na.translate=FALSE,
                    labels=c("Daily/weekly", "Every month", "More than once", "Rare", "None"), 
                    direction = -1) + 
  labs(fill = "Attacks", title = "Government Attacks on the Judiciary, 2021",
       subtitle="Based on VDem's v2jupoatck_ord Variable",
       caption="Source: VDem") +
  theme_map() + 
  theme(plot.title=element_text(face="bold", size=15, hjust=.5), 
        plot.subtitle=element_text(size=12, hjust=.5), 
        plot.caption = element_text(size=10), 
        legend.text = element_text(size = 10),
        legend.title = element_text(size = 10) )